1824
Photoelectron Spectroscopy of Azo Compounds K. N. Houk,*’* Yau-Min Chang,’* and Paul S . Engel*Ib Contribution from the Department of Chemistry, Louisiana State University, Baton Rouge, Louisiana 70803, and the Department of Chemistry, Rice University, Houston, Texas 77001, Received August 21, I974
Abstract: The photoelectron spectra of eight acyclic trans-azo compounds, cis-azoisopropane, and nine polycyclic cis-azo compounds are reported. A comparison of these spectra with those of 14 other azo compounds reported in the literature and with calculations for a variety of azo compounds allows assignments of the various ionization potentials. Correlations between the ionization potentials and quantities such as the UI of the alkyl substituents, the CNN angle, and the na* transition energy are discussed. A connection between the ground-state electronic structure of azo compounds and their propensity for loss of nitrogen upon photolysis is shown.
The properties of azo compounds (RN=NR) depend strongly on the nature of the attached R groups. At one extreme are the azo dyes, whose usefulness derives from their high stability and their electronic transitions in the visible region. When R is a relatively stable alkyl radical, the compounds are nearly colorless but are thermally labile, and they constitute an important class of polymerization initiators. The properties which make azo compounds commercially useful have also made them attractive to spectroscopists, theoreticians, and mechanistic organic chemists. Because of the many interesting chemical features of these species, summarized below, we chose to study the photelectron spectroscopy (PES) of a series of azo compounds. Our aim was to gain some insight into their ground-state and, indirectly, excited-state electronic structures. Thermal Chemistry. Depending upon the nature of the attached R group, azo compounds undergo homolysis into free radicals over a remarkably large temperature range. Azotriphenylmethane, for example, is unstable a t -40°,2a whereas azobenzene is stable even a t 600°.2bSome 45 years ago, it was recognized3 that acyclic azoalkanes could eliminate nitrogen by either one or two C-N bond scissions in the rate-determining step. While recent gas-phase studies4 have been interpreted in terms of one-bond homolysis, the solution result^^,^ are best explained as a continuous gradation between one- and two-bond scission, depending on the relative stability of the two alkyl radicals. The stereochemistry of nitrogen extrusion from cyclic azo compounds has received much attention, because even though stereospecificity of reclosure of the potential carbon diradical intermediates is generally some examples of net inversion of stereochemistry at one radical center are known.’ Nitrogen extrusions from systems in which concerted [2 2 21 cycloadditions can occur have aroused great interest, and a stereoelectronic requirement for overlap of the breaking bonds has been ider~tified.~-l’ In one case, the thermal and photochemical extrusions follow the same stereochemical course, in spite of the formal orbitalsymmetry forbiddenness of the latter process.’0%12 Photochemistry. Direct irradiation of cyclic azoalkanes generally proceeds with higher stereospecificity than does triplet sensitized p h o t ~ l y s i s . ’In ~ . ~most ~ acyclic cases, the photochemical primary process in solution is trans to cis isomerization followed by thermal homolysis of the labile cis c o m p ~ u n d Whether . ~ ~ ~ ~this ~ ~isomerization ~ involves rotation about the N N double bond or inversion a t one of the nitrogens is a subject of some i n t e r e ~ t . ~ ~ J ~ - ’ ~ Other photochemical reactions of azo compounds are hydrogen abstraction,’5,20 t a ~ t o m e r i z a t i o n , ’and ~ ionization of the a substituent.21 Electronic Structures of Azo Compounds. The basic electronic structure of the azo chromophore has been thorough-
+ +
Journal of the American Chemical Society
/
97.7
ly elucidated. Figure 1 shows the n and P orbitals of the cis (c.2”)and trans (C2h) chromophores, arranged in descending order of energy from top to bottom according to most calculations. The difference in energy between the antibonding (n-) and bonding combination (n+) of n orbitals is a sensitive function of the NNX bond angle, but in most calculations, the order of orbitals going from highest occupied down is n-, r, n+. In this work, we have chosen to use “n-” as the designation for the antibonding combination of n orbitals, whereas others have used the group theoretical point of view, where n- is the antibonding combination in CzV (cis) symmetry but the bonding combination in C2h (trans) symmetry.22 In the designation adopted here, the “n-” orbital always lies a t a higher energy than the “n+”, whether the compound is cis or trans. Theoretical Investigations of Azo Compounds. A variety of theoretical calculations on azo compounds have been reported in the literature. While a complete review of these will not be attempted, some of the pertinent calculations are summarized in Table I. Most of the recent theoretical results predict that both cis and trans azoalkanes have the n-, a, n+ order of orbital energies, where the n- orbital is the highest occupied. The only deviation from this is in cis azoalkanes, where contraction of the H N N or C N N bond angle may raise the n+ orbital energy above the r. Gimarc has analyzed BAAB molecules in general and has discussed the orbital energies of diimide as a function of geometry, using energies calculated by the extended Hiickel method.I9 Baird et al. have discussed the n- orbital energy in azoalkanes as a function of angle and have carried out a b initio calculations for cis-azomethane with C N N angles between 110 and 130°,23 while Brogli et al. have performed extended HiickeI and M I N D 0 / 2 calculations on diimide for a somewhat larger range of H N N angles.24 In these treatments, bending both HNN (or C N N ) angles from near 180’ to around 120’ results in a gradual decrease in the n - orbital energy. This is variously described as mixing of the rg orbital of linear diimide with the hydrogen 1s orbitals (Gimarc)” or as a decrease in interaction of the nitrogen lone pairs as they are bent further away from each other (Baird et al.23 and Haselbach and Heilbronner).22a As the angle is contracted down to about l l O o , there is a gradual increase in the HN or CN bonding character of the n- orbital. I n the Brogli et al. calculations, the energy of the n- orbital begins to increase again as the H N N angle is further contracted below 120’. The behavior of the n + orbital is different for the cis- and trans-diimide models. In the trans compound, the n+ orbital energy more or less parallels that of the n-, decreasing in energy from 180 to 130°, and then i n ~ r e a s i n gThe . ~ ~ n+ or-
/ April 2, 1975
1825 Table I. Summary of Selected MO Calculations on Azo Compounds Orbital energies, eV LHNN or LCNN trans-Diimide (C,h)
Diazirine
120" 100" 120" 100" 112" 39 min 9 0" 120" 150" 90" 120" 150" 112" 39 min -60"
Diazetine
-90"
Pyrazoline
-110"
cis-Diimide
(e2,>)
e (L)
-8 -10 -9.7 -9.9 - 10.46 -9 -10 -7 -9.8 -10.4 -9.0 -10.97 -10.48 -10.57 -12.85 -9.85 -12.43 -9.77 -12.29
E
(a)
e
-18 -18 -12.0 -12.2 -14.88 -18 -18 -18 -12.0 -12.0 -11.8 -15.02 -12.32 -11.1 -14.2 -11.2 -14.5 -10.4 -14.1
(nil
-20 -18 -13.3 -12.8 -17.69 -17 -17.5 -18 -11.9 -11.9 -12.0 -15.63 -14.48 -11.70 -14.02 -10.20 -12.96 -10.54 -13.50
R
Method
Ref
EHT EHT MINDO/2 MINDO/2 Ab initio EHT EHT EHT MINDO/2 MINDO/2 MINDO/2 Ab initio Ab initio MIND0/2 EHT MINDO/2 EHT MINDO/2 EHT
19 19 24 24 25 19 19 19 24 24 23 25 28 24 24 24 24 24 24
R
\
/
N=N
An
12 8 3.6 2.9 7.23 6 7.5 11 2.1 1.5 3.0 4.66 4.0 1.13 1.17 0.35 0.53 0.77 1.21
N=N
R I : Me 2: E t 3: Pr 4: A-Pr 5: 1-Bu 6: Me3CCH2CMe2-
/
R
RXR N=N
R -
R'
13: 14: 15:
U?
7: MeOCMe28: AcOCMe29: NCCMe2-
hl
IO: F 16: 17: 18: 19:
Figure 1. The n and r orbitals of the cis (Cz,) and trans (C26) azo chromophore. l h e designations in the center of the figure are used in the text.
H Me F
n I 2 3 4
n 20: I 21: 2
bital energy remains approximately constant in energy or X&/ N rises somewhat as the angle is bent from 180 to 90°, because this orbital (which would otherwise be stabilized on bending) mixes with a lower energy H N N H bonding orbitX al (which would be destabilized on bending in the absence 24 25 2 2 : CH2 of mixing). 19,24 2 3 : CH The magnitude ( A n ) of the difference in energy between the n- and n+ orbitals has been the subject of much interest. Haselbach and Heilbronner have reviewed the various N=N spectroscopic and theoretical estimates of An, which vary from 0 to 7 eV.22 Brogli et al. have calculated An for cisa n d trans-diimide as a function of the H N N angle in the 31 Z X M I N D 0 / 2 a p p r ~ x i m a t i o nSome . ~ ~ of the results are shown 29: 0 26: Me in Table I. The smallest value of An was found for cis-di2 7 : CI 30,' CMe2 imide with an H N N angle of 105' and the largest for 28: C02Me trans-diimide with an angle of 120'. Semiempirical calculations and a b initio c a l c ~ l a t i o n s ~ ~ ~ ~ ~ alike predict that An is considerably larger for trans azoalThese data are summarized in Table 11. Heilbronner's study kanes than for cis and, as a result, assuming Koopmans' of trans-azomethane established the order of orbital enert h e ~ r e r n , ~the ' trans compound should have a lower ionizagies in this prototype azo compound as n-, s,n+.22 Correlation potential than the cis, while the average of the n+ and tions with the spectra of trans-2-butene and trans-acetaldehyde methylimine, semiempirical calculations, and the obn- Zv's is lower for the cis isomers than for trans isomers. servation of vibrational structure in the ?r band led to these Previous Photoelectron Studies. Prior to and during our assignments, which have been accepted by others and used studies of the PES of azo compounds, the ionization potento make assignments in studies of more complex azo comtials of a number of azo compounds have been reported.
x'
4Q
A
Houk et al.
/
Photoelectron Spectroscopy of Azo Compounds
1826 Table 11. Previously Reported Values of Vertical Ionization Potentials (eV) of Azo Compounds
I,(U (n-) trans-Azomethane (1)
I"(2)
8.98 (8.98)
trans-Difluorodiazene (1 0) cis-Hexafluoroazomethane (12) Diazirine (1 3) Dimethyldiazirine (14) Difluorodiazirine (15) 2,3-Diazabicyclo[2.2.1] hept-2-ene (16)
1,(3) (n+)
(n)
11.84 (11.81) (1250 cm-') 14.1 15.35 13.25 12.11 15.16 11.53 11.50 10.70 10.62 10.52 11.59
14.15 13.31 -15-16.5 11.91 11.80
>12.70
2,3-Diazabicyclo [2.2.2] octene (17) 6,7-Diazabicyclo [ 3.2.21 non-6-ene (18) 7,8-Diazabicyclo [4.2.2] dec-7-ene (19) 7,8-Diazatetracyclo[ 3.3.0.0234.0326]oct-7-ene (20)
13.4 11.35 10.75 9.76 11.78 8.96 8.82 8.32 8.07 8.02 8.54
4,5-Diazatetracyclo [4.3.0.0239.03~7] non-4-ene (21)
8.65
11.64
3,4-Diaza-exo-tricyclo [4.2.1 .023']non-3-ene (22) 3,4-Diaza-exo-tricyclo[4.2. 1 . 0 2 q nona-3,7-diene (23)
8.90 9.0,
11.3, 9.4,, 11.8,
Others
12.3 (12.3)
11.2-1 1.6 12.1
Ref 22 (29)
10.41 (es), 10.95 (eas) 10.15 (es), 10.88 (eas)
10.4, 10.6,
29 29 30 31 30 32 33 32 32 32 33 33 24 24
Table 111. Vertical Ionization Potentials (eV) of Trans-Azo Comoounds (This Work) ~~
(n-) trans-Azomethane (1) (ref 20) trans-Azoethane (2) trans-Azopropane (3) trans-Azoisopropane (4) trans-Azo-tert-butane ( 5 ) trans-Azo- tprt-octane ( 6 ) trans-Azo-2-methoxyisopropane (7) trans-Azo-2-acetoxyisopropane (8) trans-Azo-2-cyanoisopropane (AIBN) (9)
8.98 8.77 8.61 8.47 8.20b 8.00 8.33 8.74 9.62
IV(2) (n)
M 3 ) (n+)
11.84 11.4, (1 1.36)a -11.1 10.83 >9.6 (10.53)a 10.9
12.3 11.7, (11.78)Q 11.5 11.2, (10.96)a
>11.5
Other I,'s
9.96 (nMeO) 10.07 ( n c o ) 10.96 (nOAC) 12.5 (nCN)
UCalculated from 01 values for the substituents (see Discussion). bReference 57.
remaining ionizations in each of these compounds. Since in pounds. Brundle and coworkers also studied trans-azomethane as well as difluorodiazene and h e x a f l u o r ~ a z o m e t h a n e , ~ ~every case this band was very broad and showed little vibrational structure, the vertical ionization potential was chosen and the assignments (Table 11) of the resolved bands agreed as the position of the maximum of a smooth curve drawn with those of Heilbronner and those inferred from a b initio through the Franck-Condon envelope. None of the azoalcalculations. The spectra of d i a ~ i r i n e ,dimethyldiazir~~ kanes 2-8 has clearly resolved ionizations assignable to the a series of bridged six-membered cyclic azo comremoval of an electron from the T or n+ orbitals so these p o u n d ~ and , ~ ~two d i a ~ e t i n e are s ~ ~of particular interest bevalues can o$y be estimated. As a guide to the identificacause of the large change in split between n- and n+ orbital tion of these bands, the position of the vertical ionization energies. Heilbronner and coworkers proposed the following potentials of azomethane are drawn a t the bottom of Figure empirical parabolic relationship between the CNN angle (e) in cis-azo compounds and the difference in energies be- 2. The ionization potentials listed in Table 111 are estimates 3.73 X tween the n- and n+ orbitals (An): An = 1.4 of unresolved maxima in the envelope arising from all ionizations below 21 eV except that of the n- ionization. The 10-3(89.5 eV.24 This function is a minimum for 0 = numbers in parentheses are calculated positions based on 89.5O, which is in good agreement with various theoretical inductive effects (see below). treatments which predict the minimum split for a diazetine, Compounds 7-9 all have lone-pair orbitals in addition to which has 0 = 90'. those on the azo group, and one or more resolved bands are The interpretation of the PES of the tetracyclic comobserved for ionizations from these orbitals. Thus, the mepounds 20 and 21 is of particular novelty.33 In the latter thoxy compound 7 has a moderately sharp band a t 9.96 eV, compound, the split between the symmetric cyclopropyl orwhich we attribute to a n ionization involving a lone pair on bital and the "" orbital is 0.31 eV greater than in the forthe methoxy group. By comparison, the lowest Z, of dimethmer, and this greater interaction is said to lead to a weaker yl ether is 9.94 eV.34 The inductive electron release by the cyclopropyl u bond and a stronger T N N bond in 21 than in two a methyls and the electron withdrawal by the a-azo 20. These differences in bond strengths are claimed to acgroup must approximately cancel in 7. In this assignment, count for the more ready thermal loss of nitrogen from 21 we neglect the fact that there should be interaction between than from 20. the lone pairs on the two oxygens and on the azo group. Photoelectron Spectra and Assignments. The PES spectra However, the fact that only a single (albeit broadened) ionreported here were recorded on a Perkin-Elmer Ltd. (Beacization band is found in the position expected for the oxygen onsfield, England) Model PS-18 photoelectron spectromelone-pair ionization indicates that the mutual through-bond ter. Resolution in every case was a t least 20 meV, and interaction of these lone pairs is not large. xenon and argon were used as internal calibrants. The phoThe acetoxy compound has two broad but resolved trantoelectron spectra of eight acyclic trans-azo compounds are sitions a t 10.01 and 10.96 eV, which can be compared with shown in Figure 2, and the vertical ionization potentials of the carbonyl and ether lone-pair ionizations in methyl acethese compounds are listed in Table 111. The lowest ionizatate, which appear a t 10.59 and 11.21 eV, r e ~ p e c t i v e l y . ~ ~ tion, which is assigned by analogy to the state arising from The broadness of these bands probably indicates that each ionization from the n- orbital, is clearly resolved from the
+
Journal of the American Chemical Society
/
97:7
/ April 2, 1975
1827
N+
NC+
//
CN
N
t
> c .v)
C 0, c
c
CI
C
2
*
c
0 0,
c .-
-W
v)
E 0)
0
c
c
0
Y
J=
a
c
2
c
V
W
W 0
c
r 0
a
N-N
I
//
8
9
I I IO
II
12
I
l
l
14
15
16
-
13
Ionization P o t e n t i a l Figure 2. Photoelectron spectra of trans-azo compounds.
band arises from two closely spaced bands so that there is some mutual interaction between the lone pairs mediated by the azo group. Finally, AIBN (9) has a partially resolved ionization at 12.5 eV, which compares favorably with the T C N band in acetonitrile, which appears at 12.2 eV.36 In all three of these compounds, the a and n+ bands are not resolved, and no attempt to find them in the u envelope has been made. The photoelectron spectra of cis-azoisopropane and the cyclic cis-azo compounds are shown in Figures 3 and 4, and the vertical ionization potentials are listed in Table IV. Azoisopropane is the first azo compound for which photoelectron spectra of both the cis and trans isomers are available. The n- ionization of the cis compound is 0.25 eV lower than that of the trans compound which indicates a larger C N N bond angle in the cis compound (see Discussion) since calculations indicate that the ionization potential of a cis azoalkane should be higher than that of the corresponding trans compound with the same C N N angle. The cis-azoisopropane molecule has two partially resolved ion-
Ionization Potential ( e V ) Figure 3. Photoelectron spectra of cis-azo compounds (24,25,27-30).
ization potentials at 1 1.18 and 1 1.9 eV, which we assign to the a and n+ ionizations, respectively. The bicyclo[2.2.1] system 16 is reported here for the third time, and the ionization potentials we find agree closely with those of Boyd et aL3* given in Table 11. The cyclic azoalkanes reported here for the first time, 24-26, all have well-resolved n- bands, but other low-energy ionizations are resolved only for the diazacyclopentene 26. The chlorodiazacyclopentene 27 has one partially resolved peak at 10.85 eV, which we assign to a chlorine lone-pair ionization (cf. ethyl chloride, I , = 11.01 eV),37 while the carbomethoxy compound 28 has a resolved band a t 10.29 eV, which we assign to the carbonyl lone pair (cf. methyl acetate 10.59 eV).35 Assignments of the bands in the PES of the compounds 29-31 all of which have two chromophores incorporated into one ring, are somewhat less straightforward than the previous assignments. The PES of compound 30 has a Houk et al.
/
Photoelectron Spectroscopy of Azo Compounds
1828 Table IV. Vertical Ionization Potentials (eV)of Cis-AzoCompounds (This Work)
(n-1 cis-Azoisopropane(1 1) 2,3-Diazabicyclo[2.2.1] hept-2ene (16) 2,3-Diaza-1,4-dimethylbicyclo [ 2.2.21 oct-2ene(24) 1,2-Diaza-3,3,6,6-tetramethylcyclohexene (25) 1,2-Diaza-3,3,5,5-tetramethylcyclopentene (26) 1,2-Diaza-3-chloro-3,5,5-trimethylcyclopentene (27) 1,2-Diaza-3-ca.rbomethoxy-3,5,5-trimethylcyclopentene (28) 3,4-Diaza-2,2,5,5-tetramethylcyclopentenone (29) 3,4-Diaza-2,2,5,5-tetramethylisopropylidenecyclopentene (30) 14,15-Diaza-7-thiadispio [ 5.1.5.21pentadec-14-ene(31)
143) (n+) 11.9 -11.9 10.93 >11 11.26
(n)
8.24 8.94 8.06 7.89 8.63 9.04 8.94 8.61 8.58 8.57
11.18 -11.5 10.48 10.82 10.91 11.24 11.18 11.53 10.72 >10.4
Other Iv's
10.85 (ncl) 10.29 (nco) 10.12 (nco) 8.61 (nee) 8.34 (ns)
10.9 11.88 11.37
Discussion Inductive Effects on Ionization Potentials. In order to discern the effect of angle distortions on the ionization potentials of azo compounds, it is first necessary to know how alkyl groups influence the ionization potentials. For the trans azoalkanes 1-5, a linear correlation between the lowest ionization potential and the polar substituent constant m4' is obtained (correlation coefficient, r = 0.996) : Zv(l) (trans azoalkanes) = (10.28 k 0.02)
+
14.12Ca1 (1) The error limits are f the standard error of estimate. The (TI'S used in this correlation are those tabulated by Levitt and coworkers, and these have been found to give excellent linear correlations with the ionization potentials of a great variety of corn pound^.^^ Particularly relevant to our discussion are the correlations found for amines:49
Ionization Potential ( e V )
Zv(l) (alkylamines) = 9.62
-
Figure 4. Photoelectron spectra of cis azo compounds (31 and 26).
broad band upon which two sharp maxima a t 8.58 and 8.74 eV are superimposed. The latter are both assigned to the alkene ?r orbital (Av = 1300 cm-'), while the n- orbital must have an unresolved maximum near 8.6 eV. The n- orbital in 30 has roughly the same ionization potential as the model 26, while the T C C ionization potential, 8.58 eV, is 0.8 eV higher than that estimated for 2,2,5,5-tetramethylisopropylidene~yclopentane.~*-~' The PES of the azo ketone 29 exhibits four resolved bands at 8.61, 10.12, 11.53, and 11.88 eV. The first band is nearly the same shape as the n- bands in the other compounds, while the second is sharper and more intense, resembling the n orbital ionization of cyclic ketones. On this basis, we assign, with some hesitation, the 8.61 eV band to the n- azo ionization. The deoxo analog 26 has the n- band at about the same position as the n- band in 29, while the carbocyclic analog 2,2,5,5-tetramethylcyclopentanone is estimated to have a lowest (no) ionization potential of 8.65 eV,42.43which is 1.5 eV lower than the no ionization in 29. The source of this large change is discussed later. Finally, the PES of the spirothiadiazoline 31 gives a band with an asymmetric shape with a maximum a t 8.34 eV. We believe that this band results from superposition of a rather sharp band at 8.34 eV and a broader, typical n- band at 8.57 eV. We can estimate crudely that the n- ionization potential of the azo analog of 31 lacking the sulfur would be at about 8.4 eV so that the S raises the n- I, to a small ( ~ 0 . 2eV) extent. The sulfur lone-pair ionization is also raised from its expected value ( ~ 7 . eV) 8 in the analogous compound lacking the azo f ~ n c t i o n . ~ ~ - ~ ~ Journal of the American Chemical Society
/
97:7
+
13.820,
(2)
By contrast to most of these correlations, the ionization potential of the parent compound, ammonia deviates significantly from the amine correlation: Ivexpt = 10.15; Ivcalcd = 9.62. A deviation in the opposite direction is observed for trans-diimide, which has a predicted I , of 10.30 eV based on 1 and an experimental value of 9.85 f 0.1 eV from electron-impact studies.50 Of additional interest are the very similar sensitivity (14.19 e V / q unit) of the azo n- ionization potential to that of amines (13.8 e V / q unit) and the greater 1, of an sp2 lone-pair azo nitrogen (calculated as 10.30 3.3/2 = 11.0 eV) vs. that of a calculated n sp3 lone pair on ammonia (9.62 eV). We have also tested the correlation between the Zv's of 23 acyclic alkylhydrazines reported by Nelsen and Buscheksl and find:
+
Z v ( l ) (alkylhydrazines) =
9.65
*
0.11
+
7.9420,
Zv(2) (alkylhydrazines) = 10.76 f 0.10 + 10.51Ca1
(Y =
0.982) (3)
(Y = 0.991)
(4)
The hydrazine nonbonding orbitals Zv's are less sensitive to alkyl substitution than are those of either alkylamines or azoalkanes. Although the P and n+ 1,'s of the trans azoalkanes are poorly resolved, we have used the values of Z,(a) for 1, 2,3, and 4 to obtain the correlations: 1,(2) (71-azoalkanes) =
(13.45 0.04) + 1 8 . 0 1 % ~ ~ (Y = 0.993) Zv(3) (n+-azoalkane) = (13.86 f 0.08) + 17.92Ca1 (Y = 0.976)
(5) (6)
Using the correlations, we calculate the values of I,(A) and Zv(n+) for trans-azopropane to be 11.36 and 11.78 eV,
/ April 2, I975
1829 Table V
respectively. These are included in Table 111. In a similar fashion, the values of I,(T) and I,(n+) for trans-azo-tertoctane ( 6 ) are calculated as 10.53 and 10.96 eV, respectively, using the value of u1 (-0.081) which is derived from eq (1) and the I,( 1) of 6. The correlation of the (TI values of 2 substituents in 2,2’disubstituted azoisopropanes 4-9 with the first Iv’sof these compounds was tested. Albeit very few substituents are involved, the alkyl and heterosubstituents give different correlations: Z”(n-1 = (8.48 f 0.03) + 3.53Cu1 = (7.03 f 0.01)
I calcd Compd
I (L)calcda
Alkyl’s
IvexPtl
13 14 16
9.40 9.15 8.37 8.46 8.25 8.20 8.16 8.30 8.29 1.96 7.96 8.13
Me, H i-PI, H i-Pr, Et PI, Et i-Pr, i-Pr i-Pr, s-Bu S-Bu, S-BU i-PI, n-Bu i-PI, Pent t-Bu, t-Bu t-Bu, t-Bu t-Bu. i-Pr
-1.35 -0.61 -0.57 -0.48 -0.07 +0.13 +0.14 -0.24 -0.36 -0.10 +0.07 -0.50
17 18 19 20 21 24 25 26
(for H, Me, neo-Pent)
Q(-N,)b
1.0
0.014 0.002 0.88 ~
+
2.12CuI
UFrom eq 1 with a 0.23 eV correction. bsummarized in ref 15.
(for OMe, OAc, CN)
The a1 values for the heterosubstituents were taken from Exner’s recent ~ o m p i l a t i o n It . ~is~ possible that, whereas the model system used for the determination of q ’ s (ionization of constants of 4-X-bicyclo[2.2.2]octane-l-carboxylic acids) measures a pure “inductive” effect of the classical variety, there is a through-bond interaction of the n or T systems on the heterosubstituent and the azo lone pair. However, the a-substituent effects on the n- and n+ Iv’s of 26-28 lie on a straight line: &(n-, 26-28) = 8.67 I&+, 26-28) = 10.94
+ +
0 . 7 5 ~ ~ (Y = 0.999) 0 . 6 1 ~ ~ (Y = 0.994)
Azoisopropane is the first compound for which PES studies of both cis and trans isomers are available. The lowest I , of cis-azoisopropane (8.24 eV) is 0.23 eV lower than that of trans-azoisopropane (8.47 eV). The calculations discussed earlier predicted a higher ionization potential for a cis azoalkane than for a trans azoalkane if the C N N angle were the same in the two compounds. The average of the n+ and n- ionization potentials in trans-diimide (1 1.5 eV by M I N D 0 / 2 a t 120’) is 0.3 eV higher than that for cis-diimide (1 1.2 eV by M I N D 0 / 2 a t 120’), but the splitting in the trans (3.6 eV) is more than twice that of the cis (1.5 eV) leading to a predicted higher ionization potential for the cis compound than for the trans. The n+ I , is not clearly resolved in the PES spectra of either the cis- or trans-azoisopropanes, so a direct comparison with calculations is not possible. However, the fair UI correlation for the estimated n+ 1,’s indicates that the estimates of I , given in Table I1 for the n+ I , are probably correct. For the cis compound, the difference between the n Iv’s may range from 2.9 to 3.7 eV depending on the assignments of the n+ ionization. The An for the cis is, therefore, equal to, or larger than that for the trans compound, in apparent contradiction to the calculations. However, an increase in the C N N angle will increase the An split. The actual C N N angle in trans-azomethane is 11 1.9°.53 For this angle in trans-diimide, the An calculated by M I N D 0 / 2 is 3.5 eV. A split of similar magnitude in cisdiimide is achieved by M I N D 0 / 2 only at a nearly 180’ angle. In our C N D O / S calculation^,^^ opening the cis-azoisopropane angle from 110 to 140’ causes an increase in the n-, n+ split from 2.1 to 3 eV. We conclude from this only that the C N N angle in cis-azoisopropane must be considerably larger than that in the trans compound, perhaps 130140’. We may estimate the effect of angle deformations on the PES of the cyclic cis azoalkanes by calculating a value for I,(n-) using eq 1 and the assumption that cis azoalkanes have 1,’s 0.23 eV less than those of the trans compounds.
This method is quite crude, but since the n+ Iv’s are not clearly resolved in the polycyclic azoalkanes, the deviation of observed from calculated n- I , is the only available handle to probe An as a function of angle. This procedure measures the bond angle deviation from that of cis-azoisopropane. The calculated values of Iv(n-), the alkyl groups used in the estimate, and the difference between the calculated and experimental I,(n-)’s are given in Table V. The crudeness of the method can be discerned from the entries for 13 and 14. However, in both cases, the experimental Iv’sare much higher than those calculated, implying a An of much less than that in cis-azoisopropane. For the compounds for which the calculated and experimental 1,’s agree fairly well (17,18,19,24, and 25), the C N N bond angles may be near that of cis-azoisopropane. The remaining compounds, 16, 20, 21, and 26 are all expected to have C N N angles contracted from those in 17 and its analogs and, since I , is larger than that calculated, An must be smaller in these molecules, as expected from theoretical considerations referred to earlier. Recalling the theoretical results at this point, the n + orbital energy in the cis compounds remains approximately constant with decreasing C N N angle, while the norbital decreases in energy until a C N N angle of about 90’ is reached.24 For all these compounds except 13 and 14, the C N N angle is greater than 90’. Compounds 29-31 are ones in which through-space interactions between the chromophores are expected. This is true, of course, for the heterosubstituted compounds discussed earlier, but the conformational freedom in those cases prevents a detailed discussion of the interactions therein. In compound 30, the K C C and P N N orbitals may interact in a through-bond fashion mediated by the intervening u bonds of T symmetry. As noted earlier, the acc ionization potential is 0.8 eV higher than calculated a model, and this seems compatible with an inductive increase of the a I , by the azo function. In ketone 29, the carbonyl n orbital and the n- orbital may interact via appropriate intervening u bonds of b2 symmetry. As noted earlier, the n- orbital is essentially unchanged from a model lacking the ketone functionality, while the ketone n I , is raised by 1.3 eV as compared with that of the compound lacking the azo group. This seems the result of an inductive raising of both ionization potentials and through-bond interaction which lowers the lowest I , and raises the second I , in this compound. Thus, the orbitals of 29 cannot be identified as pure carbonyl % or n- orbitals. Correlations between Photoelectron and Absorption Spectroscopy. Figure 5 is a plot of the A E (in eV) for the A, of the n-a* transition of azo compounds vs. the I,’s of these compounds. The absorption data used in this plot are col-
Houk et al.
/ Photoelectron Spectroscopy of Azo Compounds
1830 Table VI. The nn* Transition EneFgiegies of Azo Compounds Compd
h a , , nm
AE,eV
Solvent
Ref
trans-Azomethane (1) trans-Azoethane (2) frans-Azopropane (3) trans-Azoisopropane (4) trans-Azo-tert-butane (5) trans-Azo-tert-octane (6) frans-Azo-2-methoxyisopropane (7) tfans-Azo-2-acetoxyisopropane (8) trans-Azo-2-cyanoisopropane (9) trans-Difluorodiazene (10) cis-Azoisopropane (11) Dimethyldiazirine (14) 15 16 17 18 19 20 24 25 26 27 28 29 30 trans-Azo-2-thiomethylisopropane (32)
353 356 359 359 368 372 388 360 347 >200 382 350 (est) 336 341.5 377.5 396.0 383.5 372 385 380 355 335 326 355 326 373
3.5 1 3.48 3.45 3.45 3.37 3.33 3.20 3.44 3.57 >6.20 3.25 3.25 3.69 3,63 3.28 3.13 3.23 3.33 3.22 3.26 3.54 3.70 3.80 3.49 3.80 3.32
Hexane Gas Gas Gas Hexane Cyclohexane Ethanol Ethanol Benzene
U
b C C
a C C
c C
d
Gas
c
e
f
Hexane Hexane Hexane Hexane Cyclohexane Hexane Hexane Hexane Cyclohexane Gas Gas Hexane Hexane
K g R
K h e C C
C C
c
c C ~~
a P. S. Engel, Ph.D. Dissertation, Harvard University, 1968. b J. G. Calvert and J. N. Pitts, “Photochemistry”, Wiley, New York, N.Y., 1966. P. S. Engel and D. J. Bishop, unpublished results. d Reference 29. e Estimated from Aonset = 359 nm.31f R. A. Mitsch, J. Heterocycl. Chem., 1 , 5 9 (1964); see also ref 29. gReference 32. B. M. Trost and R. M. Cory, J. Am. Chem. Soc., 93,5572 (1971).
30.
3 25
1,
28
-.. 1’8
I 8 5
8 0
I 9 0
I 9 5
I 10
Iv,n- l e v )
Figure 5. Plot of lowest ionization potentials vs. the n r * transition in eV) of azo compounds: A = trans azoalkanes; = energies (A,,, cis and cyclic azoalkanes; 0 = heterosubstituted azo compounds.
lected in Table VI, and the Iv’s are tabulated in Tables IIIV. For the alkyl trans azoalkanes, there is a linear correlation of AE(na*) and Z,(n-):
*
(Y = 0.939) AE(nn*) = (1.90 0.02) + 0.18Iv(n_) If we assume that there is no change in J and K for the n a * transitions of compounds 1-6, then the change in a* energy upon alkylation is 0.82 that of the change in n- energy. For cis azoalkanes, Baird and coworkers have suggested that C N N angle distortions influence the n- orbital energy and the n r * transition energy to approximately the same extent.23 For the acyclic trans azoalkanes, the linear correlation of I , and A.E(na*) indicates that either the C N N angle is constant, or that it is linearly related to the q of the alkyl substituent. As seen in Figure 5, there is no general correlation of I , with AE(na*), because both alkylation and angle strain are involved in determining the n- orbital ener-
Journal of the American Chemical Society
/
97:7
gy, while, to a first approximation, alkylation, but not angle strain, will affect the a* orbital energy. There is a general trend of increasing AE(na*) with Z,(n-) for the cis and cyclic azoalkanes. If one uses compounds 11, 17, 18, 24, and 25, which are expected to have little angle distortion (see Table V), to deduce the “normal” AE(nr*) vs. I , line for cis azoalkanes, then 16 is strikingly abnormal, having a considerably larger AE(nn*) than expected. This implies an abnormally high a* energy. Similarly, 14 has too low a AE(na*) so the a* energy is abnormally low. The heterosubstituted compounds seem randomly placed, but 7, 8, 27, 28, and 29 are located near a line of high slope, indicating that the a* energy is affected by the heterosubstituent much more than the n- level. Compound 30 has a much higher AE than expected on this basis, which implies that the x*” orbital is destabilized by mixing with the r*cc orbital. Ground State Electronic Structures and Chemical Reactivity. In previous sections, we have analyzed the photoelectron spectra of a variety of azo compounds using familiar concepts of inductive effects and strain and, assuming Koopmans’ theorem, we have discussed ground-state orbital energies. Although there is no direct relationship between orbital energies and chemical reactivity, an analysis of orbital energies and the nature of each orbital may lead to some insights into the differences in the strengths of various bonds in two similar molecules. Schweig, Trost, and their coworkers have reasoned in this fashion for the compounds 20 and 21.33 They proposed that the lower the ionization potential for one of the orbitals heavily localized on cyclopropane, the easier it is to break the cyclopropyl bond, while the higher the ionization potential for the a and n+ orbitals of the azo group, the more fully formed the N2 triple bond, and the easier it is to split nitrogen from a molecule thermally. A theoretical justification for arguments of this type is not straightforward since the bond order of a given bond will be the sum of partial bond orders for all the molecular orbitals of the molecule in question. However, other empiri-
/ April 2, 1975
cal observations of this type may be made. Thus, for the series of azoalkanes 4,5, and 6, the relative rates of thermolytic loss of nitrogen are 1, 5.6 X lo2, and 7.2 X IO5, respect i ~ e l y the , ~ ~same as the order of decreasing ionization potential. While it might be argued that the lowering of the n-, T, and n+ ionization potentials reflects the decreased CN bonding in these orbitals as the hyperconjugative interaction increases, the lower energy u orbitals of the alkyl group will become increasingly stabilized, with a concomitant CN bond strength increase along this series. Furthermore, for a more general series of a z o a l k a n e ~there ~ ~ is no correlation between ionization potential and rate of loss of nitrogen. Another application of molecular orbital arguments to azoalkanes has been given by Inagaki and Fukui, who have applied frontier molecular orbital theory to the description of loss of nitrogen from azoalkanes. They suggest that as the CN bonds of azoalkanes break, the mixing of the lowest unoccupied orbital of the u framework with the highest occupied orbital of the azo H system will occur.12 However, photoelectron studies show that the n- orbital is clearly the highest occupied orbital, and the azo H* orbital is undoubtedly the lowest vacant. Since the n- orbital is also a C N bonding orbital (antisymmetric with respect to a plane bisecting the NN linkage), then the favored rotation of the groups bonded to nitrogen will be disrotatory. A stepwise mechanism involving cleavage of a single CN bond seems better able to account for the single inversion stereochemistry sometimes observed. A more promising case of photoelectron spectroscopic data is in the rationalization of photochemical reaction rates. Some success in this endeavor has already been reali ~ e d . ~ ~ Upon excitation of a molecule from ground to a n excited state, to a first approximation, bonding changes are dominated by the difference in bond orders in the orbital vacated and the orbital newly occupied. Information about the vacated orbital can be inferred from photoelectron spectroscopy in conjunction with calculations. For azoalkanes, nH* excitation causes loss of nitrogen from those molecules which cannot undergo &,trans isomerization. If we assume that the quantum yield of azoalkane photolysis depends more on the rate of cleavage of the CN bond than it does on the rate of various deactivation processes, it follows that a lower CN bond strength should lead to a higher quantum yield. For compounds 16, 26, 24, and 25, the quantum yields for loss of nitrogen (Table V) are 1.0, 0.88, 0.014, and 0.002.15The lowest ionization potentials for these compounds are 8.94, 8.63, 8.06, and 7.89 eV; that is, as the ionization potential decreases in this series, the quantum yield for loss of nitrogen decreases. Of more relevance, the deviations from calculated ionization potentials based on inductive effects are -0.57 (or -0.48), -0.50, -0.10, +0.07, as listed in Table V . The compounds whose ionization potentials are too high (16 and 26) lose nitrogen readily, while those with normal ionization potentials (24 and 25) do so with low quantum yields. As noted by Gimarc, the decrease in the n- orbital energy as the H N N angle of diimide is decreased from 180' to smaller angles is a result of mixing with the hydrogen 1s orbitals in a bonding f a ~ h i 0 n . lIn ~ C N D O / S calculations on cis-diimide, performed here, the n- orbital electron density on the hydrogens increases from 10% a t H N N angles of 150' to 26% a t 90'. The deviations from calculated ionization potentials noted in Table V are indirect measures of how much the C N N angle is constricted below that in cis-azoisopropane and, using the calculational results noted above as a guide, how much CN bonding character there is in the n- orbital. When there is more CN bonding character in the n- orbital, then the CN bond
-
-
1831
is weakened more upon nH* excitation. Admittedly, there are too few points available to show a general correlation, but we believe that this type of argument portends the power of photoelectron spectroscopy in the explanation of diverse photochemical phenomena. Acknowledgment. Acknowledgment is made to the National Science Foundation, the National Institutes of Health, and the Camille and Henry Dreyfus Foundation for financial support of this research. We thank Professor J. W. Timberlake for generously supplying samples of 6-8 and Linda S. Lambert (LSU) for recording some of the photoelectron spectra. References and Notes (1) (a) Louisiana State University; (b) Rice University, (2) (a) M. T. Jaquis and M. Szwarc, Nature (London), 170, 312 (1952). (b) H. Weiland. H. Hove, and K. Borner. Justus Liebigs Ann. Chem., 446, 31 119261
H. C. Ramsperger, J. Am. Chem. SOC.,51, 2134 (1929). R. J. Crawford and K. Takagi, J. Ann. Chem. SOC.,94, 7406 (1972), and earlier papers. S. Seltzer and F. T. Dunne, J. Ann. Chem. SOC.,87, 2628 (1965); J. Him, A. Oberlinner, and C. Ruchardt. Tetrahedron Lett., 1975 (1973); W. A. Pryor and K. Smith, J. Am. Chem. SOC.,69, 1741 (1967); 92, 5403 (1970); R. A. Johnson and S.Seltzer. ibid., 95, 938 (1973). N. A. Porter and L. J. Marnett, J. Am. Chem. SOC.,95, 4361 (1973), and earlier papers. H. Tanida. S. Teratake, Y. Hata, and M. Watanabe, Tetrahedron Lett., 5341 (1969); P. B. Condid and R. G. Bergman, Chem. Commun., 4 (1971); D. F. Eaton, R. G. Bergman. and G. S. Hammond, J. Am. Chem. SOC..94, 1351 (1972). T. Van Auken and K. L. Rinehart, Jr., J. Am. Chem. SOC., 84, 3736 (1962); D. E. McGreer, R. S.McDaniel, and M. G. Vlnje, Can. J. Chem., 43, 1389 (1965); C. A. Overberger and J. W. Stoddard, J. Am. Chem. SOC., 92, 4922 (1970); J. W. Tirnberiake and B. K. Bandlish, Tetrahedron Lett., 1395 (1971). M. Martin and W. R. Roth, Chem. Eer., 102, 811 (1969); N. Rieber, J. Alberts, J. A. Lipsky, and D. M. Lernai, J. Am. Chem. SOC., 91, 5668 (1969); J. A. Berson and S.S.Olin. ibid., 92, 1086 (1970). J. A. Berson and S.S.Olin, J. Am. Chem. Soc., 92, 1086 (1970). H. Tanida and S. Teratake, Tetrahedron Lett., 4991 (1970); E. L. Allred and K. J. Voorhees, J. Am. Chem. SOC.,95, 620 (1973); L. A. Paquette and M. J. Epstein, ibid., 95, 6717 (1973). S.lnagaki and K. Fukui, BUN. Chem. SOC.Jpn., 45,824 (1972). For a summary of much of the work in this area, see P. S. Engel and B. M. Monroe, Adv. Photochem., 6, 281 (1971). T. Sanjiki, M. Ohtu, and H. Kato, Chem. Commun., 638 (1969); R . Moore, A. Mishra, and R. J. Crawford. Can. J. Chem., 46, 3305 (1968). P. S.Engei and C. Steel, ACC.Chem. Res., 6, 275 (1973). T. Mill and R. S. Stringham, Tetrahedron Lett., 1853 (1969). J. Alster and L. A. Burnelle, J. Am. Chem. SOC.,89, 1261 (1967); J. M. Lehn and B. Munsch, Theor. Chim. Acta, 12, 91 (1968); M. S.Gordon and H. Fischer, J. Am. Chem. SOC.,90, 2471 (1968); N. C. Baird and J. R. Swenson, Can. J. Chem., 51,3097 (1973). D. R. Kearns, J. Phys. Chem., 69, 962 (1965). B. M. Girnarc, J. Am. Chem. SOC.,92, 266 (1970). See V. Malatesta and K. U. Ingold, Tetrahedron Lett., 331 1 (1973). N. Levi and D. S.Maiament, lsr. J. Chem., in press. (a) E. Haselbach and E. Heilbronner, Helv. Chem. Acta, 53, 684 (1970); (b) E. Haselbach and S.Schrnelzer. ibid., 54, 1575 (1971); (c) ibid.. 55, 1745 (1972). N. C. Baird, P. DeMayo, J. R. Swenson. and M. C. Usselman, J. Chem. Soc.,Chem. Commun., 314 (1973). and references therein. F. Brogli, W. Eberbach, E. Haselbach. E. Heilbronner, V. Hornung, and D. M. Lemai, Helv. Chirn. Acta, 56, 1933 (1973). M. B. Robin, R . R. Hart, and N. A. Kuebler, J. Am. Chem. SOC.,89, 1564 (1967). R. Ditchfield, J. E. Dei Bene, and J. A. Pople, J. Am. Chem. SOC.,94, 703 (1972). T. Kooprnans, Physica (Utrecht), 1, 104 (1934). M. E. Robin. H. Easch, N. A. Keubler, K. 8. Wiberg, and G. B. Ellison, J. Chem. Phys., 51, 45 (1969). C. R. Brundle, M. B. Robin, N. S. Kuebler, and H. Basch, J. Am. Chem. SOC.,94, 1451 (1972). M. B. Robin, C. R. Brundle, n. A. Keubler, G. 8. Ellison, and K. B. Wiberg, J. Chem. Phys., 57, 1758 (1972). E. Haselbach, E. Heiibronner, A. Mannschreck, and W. Seitz. Angew. Chem., Int. Ed. Engl., 9, 902 (1970). R. J. Boyd. J. C. Bunzli, J. P. Snyder, and M. L. Heyman, J. Am. Chem. SOC.,95, 6478 (1973). H. Schmidt, A. Schweig. 8. M. Trost. H. 8. Neubold, and P. H. Scudder, J. Am. Chem. SOC.,96, 622 (1974). B. J. Cocksey, J. H. D. Eland, and C. J. Danby, J. Chem. SOC. 6,790 (1971). R. Sustmann and H. Trill. Tetrahedron Len., 4271 (1962). R. F. Lake and H. Thompson, Proc. R. SOC. London, Ser. A, 317, 187 (1970). K. Kimura, S. Katsumata, Y. Achiba, H. Matsumoto, and S. Nagakura.
Houk et al.
/ Photoelectron Spectroscopy of Azo Compounds
1832 Bull. Chem. SOC.Jpn., 46, 373 (1973). (38) A crude estimate of the ionization potential of 2,2.5,5-tetramethylisopropylidenecyclopentane (32) can be made in the following way. Since the L, of methylenecyclopentane (8.94 eV)38 Is 0.27 eV lower than that of isobutylene (9.23 eV).'O the .,/ of propyliienecyclopentane should be about 0.27 eV lower than that of 2,MimethyC2-butene (8.30 eV)" or 8.03 eV. Furthermore a meth lation of 1-butene causes a 0.07 eV decrease in ionization potentia? so that of 32 should be: 8.03 4(0.07) = 7.8 eV. (39) D. A. Demeo and M. A. El-Sayed, J. Chem. Phys., 32, 2622 (1970). (40) K. Watanabe, T. Nakayama, and J. Mottl, J. Qoant. Spectrosc. Radht. Transfer, 2, 369 (1962). (41) R. Bralsford, P. V. Harris, and W. C. Price, Proc. R. Soc. London. Ser. A, 256, 459 (1960). (42) Di-tert-butyl ketone has an L, (8.71 eV)34 which is 0.4 eV lower than that of diethyl ketone (8.31 eV),34 and assuming a similar difference between cyclopentanone (9.25 eV)43 and tetramethylcyclopentanone gives an estimate of 8.85 eV for the I, of the latter. (43) D. Chadwick, D. C. Frost, and L. Weiler, Tetrahedron Lett., 4543 (19711. (44) Dimethyl sulfide (L, = 6.44 eV),45 di-tert-butyl sulfide (L, = 6.07 eV), and tetramethylene sulfide (L, = 8.40 eV)46are compounds used in this crude estimate. (45) H. Bock and G. Wagner, Angew. Chem., lnt. Ed. Engl., 11, 150 (1972).
I
-
(46) J. C. Bunzli, D. C. Frost, and L. Weiler. J. Am. Chem. SOC., 95, 7880 (1973). (47) R. W. Taft, Jr.. and I. C. Lewis, Tetrahedron, 5, 210 (1959). (48) H. F. Widing and L. S. Levitt. Tetrahedron, 30, 611 (1974), and references therein. (49) B. W. LevittandL. S.Levitt, lsr. J. Chem., 9, 71 (1971). (50)S. N. Foner and R. L. Hudson, J. Chem. Phys., 28, 719 (1958). (51) S. F. Nelsen and J. M. Buschek, J. Am. Chem. Soc., 96, 2392 (1974). (52) 0. Exner in "Advances in Linear Free Energy Relationships", N. B. Chapman and J. Shorter, Ed., Plenum Publishing Co.. London, 1972, p 1. (53)C. H. Chang, R. F. Porter, and S. H. Bauer, J. Am. Chem. SOC., 92, 5313 (1970). (54) J. E. Del Bene and H. H. Jaffe, J. Chem. Phys., 48, 1807. 4050 (1968); 49, 1221 (1968); 50, 563 (1969). (55) J. W. Timberlake, A. W. Garner, and M. L. Hodges, Tetrahedron Lett., 309 (1973). (56) K.-W. Shen and N. A. Kuebler, Tetrahedron Lett., 2145 (1973); R . Gleiter, E. Heilbronner, M. Hekman, and H.-D. Martin, Chem. Ber., 106, 28 (1973); W. Schmidt and B. T. Wilkins. Tetrahedron, 28, 5649 (1972). (57) NOTE ADOEDIN PROOF: T. Koenig, T. Balle, and W. Snell, J. Am. Chem. SOC.,97, 662 (1975), reported 8.33 eV for the lv of azo-tert-butane. The spectra are in other respects identical.
Thione Photochemistry. Photoreduction of Adamantanethione' ,2 J. R. Bolton, K. S. Chen, A. H. Lawrence, and P. de Mayo* Contributionf r o m the Photochemistry Unit, Department of Chemistry, University of Western Ontario, London, Canada. Received July 15, 1974
Abstract: Excitation of adamantanethione in the n,r* band in the presence of the corresponding thiol gives diadamant-2-yl disulfide as the sole product. A less efficient thermal reaction also occurs to give the same product. Use of the a-deuterated thiol results in the incorporation of 98% of one atom of deuterium in the disulfide and the requirement that there can be a chain process. This conclusion is supported by quantum yield measurements. A mechanism is proposed involving a number of radical intermediates. These intermediates have been detected and identified by utilizing the electron spin resonance "spin trapping" technique. The efficiency of free-radical trapping by thiones is discussed.
The first report of the photoreduction of a thione appears to be that of Oster, Citarel, and G ~ o d r n a nThese .~ authors showed that excitation of thiobenzophenone into S2 (using light of 366 nm) in ethanolic solution gave reduction products. These included dibenzhydryl disulfide, benzhydryl mercaptan, and a substance said to be bis(benzhydry1dithi0)diphenylmethane. In contrast, no reduction was said to occur upon excitation into S I (n,a*, A,, 599 nm). Reinvestigation of this reaction4 more recently has shown that other products are formed in addition, and that a similar, but not identical, mixture is obtained a t both wavelengths. The process occurring upon excitation to S1 was significantly slower than that a t S2. It has also been established that the thiobenzophenone 3(n,x*) state is reduced by 2-propan01,~but it has not been verified that excitation into S2 results in reaction from the 3 ( n , ~ * state ) only. This is necessary since it has been established that the S2 state of thiobenzophenone has a long enough lifetime for bimolecular reaction.6 Finally, the photoreduction of thiobenzophenone in ether (A >300 nm) has been shown to involve radicals of the type Ar2CSR, the presence of which has been detected by ESR spectroscopy.' When the present studies of the possible photoreduction of the saturated thione function were undertaken,8 only the first report on thiobenzophenone reduction was available. We report here our results concerning the long wavelength ( n , a * ) photoreduction of adamantanethione, which indicate a related mechanism to be involved, and describe the spin trapping of the intermediate species. Journal of the American Chemical Society
/ 97:7 /
Experimental Section Materials. Adamantanethione, 2-adamantanethiol, and di( 2adamantyl) disulfide were prepared according to the methods described by G r e i d a n ~ s .2-Deuterio-2-adamantanethiol ~ was prepared by reduction of the thione with sodium borodeuteride. It contains 93.5% deuterium as estimated mass spectrometrically on the corresponding disulfide. 4,4'-Dimethoxythiobenzophenone was synthesized by the method of Scheeren, Ooms, and Nivard.Io The nitroso-tert-butane was prepared by oxidation of tert-butylhydroxylamine with aqueous alkaline hypobromite solution' I and had mp 8 0 - 8 1 O (lit.II 83-84O). Benzene was purified by the method of Kraus and Vingee.]* All chemicals (including solvents) were distilled or vacuum sublimed prior to use. ESR Measurements. E S R spectra were obtained on a Varian E12 electron spin resonance spectrometer. A 100-kHz modulation frequency and -2-3 m W microwave power were employed in the detection system. ESR signal from Mn2+ in S r O was used for hyperfine splitting ~ a l i b r a t i o n . The ' ~ g factor of DPPHI4 (2.0037 I 0.0002) was used as a reference for g-factor measurements. The light source was a Hanovia Model 977B-1 I-kW Hg-Xe lamp in a Schoeffel Model LH l 5 l n lamp housing. For given wavelength limits, Corning glass color filters were employed. The solutions for photolytic E S R measurement were prepared by vacuum transfer, the purified and separately degassed benzene being distilled into the previously degassed reactants (four freeze-thaw cycles; residummHg). al pressure 5 X Photochemical Procedures. Solutions were prepared in a manner similar to that described for ESR solutions. Concentrations of thione were normally 0.05-0. I M and those of the thiol 0.5-1 .O M . Irradiations were carried out in two ways: (a) on an optical bench using a 150-W xenon Hanovia compact arc lamp and a Bausch
April 2, 1975